Reducing the Risk of Nuclear War Taking Nuclear Weapons Off High Alert

Twenty-five years after the end of the Cold War, the and continue to keep many hundreds of nuclear weapons on high alert, ready to be launched in minutes. This alert status—frequently called “hair-trigger alert,” “” status, or other synonyms1—allows both countries to launch missiles quickly in response to warning of an incoming nuclear attack, before the attacking missiles hit their targets. This Cold War–era policy is dangerous because it increases the chance of an accidental, unauthorized, or mistaken launch of nuclear weapons, as historical examples of false warning and other mistakes demonstrate. Growing tensions between the United States and Russia make it even more important to ensure that should a crisis develop—increasing the time pressure on decision makers and opportunities for misunderstandings—high alert status does not lead to a mistake that sparks an unintended nuclear exchange. In addition, the Chinese military has recently begun to argue that should put its nuclear weapons on alert for the first time and build an early warn- ing system to detect an incoming attack. These steps would also increase the risk U.S. Air Force/Josh Aycock

One of 15 missile launch control centers at Malmstrom Air Force Base in Montana. Under high alert, military personnel have only three or four minutes to decide if warnings of foreign launches are genuine.

1 The U.S. military refers to this status as “high alert,” “ready alert,” “day-to-day alert,” “launch under attack” status, or “prompt-launch” status. of mistaken launches, especially since history shows that false The Cold War Origins of High Alert alarms are more common early in the operation of a new warning system. U.S. nuclear policy is the largest external in- The policy of keeping nuclear weapons on high alert is fluence on Chinese nuclear policy. As a result, a U.S. decision a holdover from the Cold War. Both the United States and to eliminate prompt-launch options from its nuclear war Soviet Union feared the other country might launch a first plans and take its missiles off high alert could affect the inter- strike attack consisting of thousands of nuclear weapons in nal Chinese debate and help influence China to refrain from an attempt to destroy bombers and missiles before they could taking these dangerous steps (Kulacki 2015a). be launched. At that time, leaders in both nations were con- For these reasons, the United States should remove op- cerned about the vulnerability of their nuclear weapons and tions from its nuclear plans for launching missiles on warning command and control systems to such an attack. To make of an attack, and take its silo-based missiles off alert. These clear it could launch a devastating nuclear response even fixed land-based missiles are vulnerable to attack and there- after a first strike, each country kept some of its weapons fore U.S. decision makers would be under the most pressure on high alert so they could be launched in retaliation within to launch them on warning. The United States also keeps sub- minutes of an attack being detected rather than being marine-based missiles at sea on high alert, but these are not destroyed on the ground. vulnerable to attack so there would not be such pressure to launch them on warning. It is important that the United States remove rapid- The United States should launch options from its nuclear plans. This change would mean that it would not return its missiles to high alert status remove options from its in a time of crisis. nuclear plans for While it would be preferable for both the United States and Russia to take these steps, the United States should not launching missiles on wait for Russia to act. Taking U.S. land-based missiles off high alert and removing rapid-launch options from U.S. nuclear warning, and take its silo- plans would still reduce the risk of nuclear use and thereby based missiles off alert. increase national and international security. At the same time, the United States would maintain a robust deterrent against nuclear attack. In 1961, the United States began keeping nuclear-armed bombers in the air 24 hours a day so they could not be destroyed on the ground. It ended this practice in 1968 after several accidents involving the aircraft and their nuclear weapons made clear the dangers of this policy (Department of Defense BOX 1. 1981). The final straw was the 1968 crash in Thule, Greenland, Mistaken, Accidental, and of a U.S. bomber on airborne alert carrying four nuclear bombs, due to a fire on the aircraft. The conventional high Unauthorized Launches explosives in the bombs detonated, which dispersed pluto- nium over a large area surrounding the wreck (Sagan 1993). Under U.S. policy, only the U.S. president can order the The United States developed a system of early warning use of nuclear weapons. If the president is incapacitated, radars in the late 1950s that could detect attacking missiles there is a chain of command that would be followed. An early in their flight and provide 15 minutes or more of warn- unauthorized launch is a deliberate launch that would take ing of an attack. As a result, following the Thule accident, the place without a presidential order. It could be perpetrated United States replaced its airborne bombers with bombers by insiders (e.g., ordered by those in the chain of command with access to the launch codes) or by outsiders (e.g., kept on alert on the ground—armed and sitting on the runway through a cyberattack). A mistaken launch would be autho- ready to take off on warning of attack. rized by the president, but in response to a false warning Along with the development of long-range missiles that of an incoming attack. An accidental launch would not could be launched quickly—in particular the Minuteman and be deliberate, but would occur through a system error Titan II missiles—the early warning system also allowed the (e.g., a computer glitch). United States to put land-based missiles on alert starting in 1962. But the capability to quickly launch missiles came with

2 union of concerned scientists The United States and Russia each deploys 1,800 to 2,000 nuclear weapons, and each maintains roughly 900 missile- launched nuclear weapons on high alert. U.S. Air Force U.S. In another close call incident in 1961, two nuclear bombs fell to the ground in North Carolina when a bomber lost a wing. Neither bomb detonated, but multiple submarines at sea, which could move into range of their safety devices failed. According to Defense Secretary Robert McNamara, “By the slightest margin of chance, literally the failure of two wires to cross, a nuclear targets and be on alert within days. Most of Russia’s nuclear explosion was averted.” warheads on high alert are believed to be on its silo-based missiles (Kristensen and McKinzie 2012). a price. Unlike bombers, missiles cannot be called back or Both the United States and Russia have options in re-targeted after launch. Nor do they carry self-destruct their nuclear war plans for launching missiles quickly if they mechanisms to abort a mistaken launch. Once fired, the receive warning of an incoming nuclear first strike. Indeed, missiles will proceed to their targets. those options drive the requirement to keep missiles on Unlike the United States, the Soviet Union did not rou- high alert. tinely keep armed strategic bombers on alert—either airborne Britain and maintain SLBMs at sea, but they are or on the ground. And it did not have a warning system to thought to be on lower levels of alert than U.S. and Russian allow it to place its missiles on alert until the early 1970s missiles. China currently keeps its nuclear warheads stored (Podvig 2001). separately from its missiles, and therefore its forces are com- In 1991, following the end of the Cold War and with pletely off alert. The other nuclear-armed nations—, relations warming between the United States and Russia, , , and —also keep their weapons President George H.W. Bush ordered U.S. nuclear bombers off alert, with their warheads stored separately from their taken off alert, with their nuclear weapons stored separately delivery systems. from the bombers. These planes are no longer ready to take off within 15 minutes, but can still take off within 24 hours. Yet both countries continued to keep nuclear missiles BOX 2. on high alert. Launch Under Attack The Current Situation While launching quickly on warning of attack is typically The United States and Russia each deploys 1,800 to 2,000 called “launch on warning,” the United States instead calls nuclear weapons, and each maintains roughly 900 missile- such use “launch under attack.” It states that launch under attack is based on an “attack assessment that considers launched nuclear weapons on high alert. For the United and confirms warning information from multiple, inde- States, the alert forces include all but a few of its 450 silo-based pendent sensors” and “also considers the apparent intent intercontinental ballistic missiles (ICBMs), each armed with of the incoming attack in the context of the international one warhead, and a comparable number of warheads on situation” (Department of State 2015a). The implication submarine-launched ballistic missiles (SLBMs) on the four is that the United States would launch only if it knew that to five submarines that are kept within range of their targets. an actual nuclear attack was under way—but this is The ICBMs can be launched within a couple minutes of misleading since such confidence is not possible until a presidential decision to do so, and the SLBMs within nuclear explosions are detected. 15 minutes. The United States keeps four to five additional

Reducing the Risk of Nuclear War 3 The Dangers of Keeping Missiles nuclear response. Short decision time is especially problem- on High Alert atic when there is confusion and ambiguous information, and at times of high international tensions. The current U.S. policy of keeping missiles on high alert in- It takes about 30 minutes for a missile to fly between creases the risk that one or more would be launched by acci- the United States and Russia. In the case of an attack by dent, without authorization, or in response to a false warning submarine-launched missiles, the flight time could be of an incoming attack. Any such launch would be catastrophic much shorter. After a missile is launched, it takes time for and could in turn trigger a nuclear attack by Russia. a country’s early warning system of space-based and ground- In engineered systems as complex as the U.S. and Russian based sensors to detect the attack, leaving even less time early warning and nuclear command and control systems, available to make decisions before an attack lands. The it is common for unanticipated behaviors and failure modes timeline in place for responding to warnings is therefore to arise as a result of both technical glitches and inevitable very compressed. human errors (Perrow 1984). The unexpected nature of such events can make it difficult to determine whether warnings are real, especially within the few minutes allowed by launch- In the United States, on-warning options. Assessing warning can be further compli- cated by coincidental but unrelated events, as well as by the president would have multiple events that appear to be independent of one another at most 12 minutes to but share a common underlying cause. As discussed below, numerous incidents of false warnings and other problems decide whether and how have occurred in both countries. to respond with nuclear The fact that no accidental, unauthorized, or mistaken launches have occurred so far suggests that the safety measures weapons. put in place by the two countries work well enough that the probability of accidents and errors leading to a nuclear launch is small. In the United States, which has a very advanced But the probability is not zero. And the more of these early warning system, the president would have at most incidents that occur, the greater is the chance that, due to 12 minutes—and likely much less—to decide whether and confusion and an unforeseen confluence of events, one of how to respond with nuclear weapons (Global Zero 2015). them will lead to disaster. The U.S. early warning system would detect a missile within It is worth remembering the explosion of the space three minutes of its launch; the personnel monitoring the shuttle Challenger in January 1986. Before the accident, NASA systems would have no more than three or four minutes to management claimed the probability of a catastrophic acci- determine whether the warning was genuine and to pass in- dent was about 1 in 100,000. Afterward, analysis showed that formation up the chain of command. Senior military leaders the probability was instead closer to 1 in 100, as had been would confer and then quickly brief the president and present maintained all along by engineers in the program (Feynman him or her with options for a response. This briefing would 1986). This figure proved to be consistent with the actual rate last no more than a minute, and the president would then of accidents that occurred: there were two fatal accidents— have to make a decision in time to transmit launch orders to Challenger in 1986 and Columbia in 2003—out of 135 total launch crews before the incoming missiles hit (Starr n.d.). flights. Yet the original unrealistic estimate of the probability History has shown, however, that the process may not shaped the thinking about risks in the program. Similarly, fit this rushed timeline. In two past incidents involving false the actual probability of a catastrophic nuclear event may alarms in the United States, the personnel at the warning in fact be larger than is typically assumed. centers reportedly took eight minutes to sort through the confusing data they were receiving. In both cases, this led

SHORT DECISION TIME IS DANGEROUS to them being relieved of duty (Global Zero 2015). Even if the ultimate decision is not to respond immedi- The underlying problem with retaining options to launch ately, maintaining the option of launching weapons before an on warning of attack is the short time available to assess the attack lands still increases the pressure on military officers warning, to determine whether there is an actual attack and and decision makers to act quickly. Intense time pressure whether it is nuclear, and to decide whether to launch a is not helpful for good decision making, especially in the

4 union of concerned scientists context of ambiguous or conflicting information. Moreover, compressed and the opportunities for ill-considered decisions if warning occurs during a crisis, that context may predispose very real. Launch-on-warning puts enormous strain on the those assessing the warning to see it as credible. nuclear chains of command in both countries” (Cartwright Retired U.S. and Russian Generals James Cartwright and Dvorkin 2015). and Vladimir Dvorkin, both of whom have extensive experi- The short decision timeline has required the militaries ence with nuclear weapons,2 have warned about this prob- of both countries to develop and practice nuclear launch pro- lem. They write, “for either side, these timelines are very cedures that can be carried out quickly and routinely. General

BOX 3. U.S. Military Experts Endorse Taking Nuclear Weapons Off High Alert

General James Cartwright (Marine Corps four-star general, Senator Sam Nunn (U.S. senator from Georgia and chair vice chairman of the Joint Chiefs of Staff, and commander of the U.S. Senate Armed Services Committee): of U.S. Strategic Command) chaired a May 2012 report that I do not believe that our continued Cold War operational stated: status adds to our deterrence or enhances either side’s security; The current postures of launch-ready nuclear forces that it does, however, increase the chance of a catastrophic provide minutes and seconds of warning and decision time accident made from too little information and too little should be replaced by postures that allow 24–72 hours on time. . . . Both sides could increase decision time by elimi- which to assess threats and exercise national direction over nating the prompt launch readiness requirement for as the employment of nuclear forces. This change would great- many forces as possible, getting these weapons off hair ly reduce the risks of mistaken, ill-considered and accidental trigger (Nunn 2002). launch (Global Zero 2012). General William Odom (Army three-star general and General Eugene E. Habiger (Air Force four-star general and director of the National Security Agency): commander of U.S. Strategic Command): I don’t see why we have the forces alert. I’ve never been We have to find a way to move more nuclear weapons a big enthusiast for our whole approach of being able to off alert status and give leaders more decision time in a launch on warning or launch in a very short amount of crisis (Habiger 2002). time. Firing off 1,000 or 500 or 2,000 nuclear warheads on a few minutes’ consideration has always struck me as an Robert S. McNamara (U.S. secretary of defense): absurd way to go to war. . . . Therefore I think it would make a lot of sense to completely de-alert (Frontline 1999a). The risk of an accidental or inadvertent nuclear launch is unacceptably high. . . . At a minimum, we should remove all Stansfield Turner (Navy admiral and director strategic nuclear weapons from ‘hair-trigger’ alert, as others of the Central Intelligence Agency): have recommended, including Gen. George Lee Butler, the last commander of SAC. That simple change would greatly I think that one of the first things we should do is take every reduce the risk of an accidental nuclear launch. . . . U.S. weapon off of high alert. We have an absolutely insane policy in this country. Had it now for 30 or 40 years. . . . The indefinite combination of human fallibility and nuclear Our missiles that count are in submarines out here at sea, weapons carries a very high risk of nuclear catastrophe. and they can’t see those. So we can always counterattack, There is no way to reduce the risk to acceptable levels, no matter what they do in that attack (Frontline 1999b). other than to first eliminate the hair-trigger alert policy and later to eliminate or nearly eliminate nuclear weapons (McNamara 2009).

2 James Cartwright is a retired Marine Corps general and former vice chairman of the Joint Chiefs of Staff (2007–2011) and commander of the United States Strategic Command (2004–2007). Vladimir Dvorkin is a retired major general and former head of the research institute of Russia’s Strategic Rocket Forces (1993–2001).

Reducing the Risk of Nuclear War 5 Cartwright describes this as “gearing the nuclear command- control-communications and warning system from the Military officers with president on down to the individual launch commanders experience overseeing U.S. for rapidly executing the forces in the opening phase of a nuclear conflict” (Cartwright 2013). nuclear forces are concerned Similarly, a 2015 report by more than two dozen retired and active military officers and security experts states that, about the vulnerability to “During the Cold War both sides honed procedures to send cyberattack of missiles on the go-code at the first signs of incoming warheads reported by early warning satellites and ground radar. Under this plan high alert. to launch on warning, which remains intact and frequently exercised on both sides today, nuclear decision-making is extremely rushed and emotionally charged. To prevent panic, Global Zero Commission on Nuclear Risk Reduction, chaired it is pre-scripted, driven by checklists, and enacted by rote” by General Cartwright, notes that missiles are “ready to (Global Zero 2015). launch upon receipt of a short stream of computers signals” According to some military officers who have taken part and concludes that: in the procedure, these rapid-reaction procedures have the [G]iven so many unanswered questions and our weak effect of biasing the process toward a decision to launch. comprehension of this cyber threat, we have yet another General Lee Butler, who from 1992 to 1994 was commander of reason for concern about strategic missiles on high alert Strategic Air Command, which oversaw nuclear bombers and and about trends among the other nuclear weapons coun- land-based missiles, has said the Cold War view of needing tries toward increased attack readiness of their nuclear to launch quickly is still built into U.S. nuclear policy at an forces. If we cannot fully assess the risks, it would seem operational level. He states that nuclear planners “built a con- prudent to keep nuclear missiles off of high alert status struct that powerfully biased the president’s decision process at all times. This would be a sure-fire way to mitigate toward launch before the arrival of the first enemy warhead . . . foreseeable risks as well as those that have not yet a move in practice to a system structured to drive the presi- been imagined (Global Zero 2015). dent invariably toward a decision to launch under attack,” and that this dangerous prejudice continues to guide the The report discusses, for example, a potential vulnerability process today (Blair 2015). of the command and control system of U.S. ICBMs: someone could tap into the thousands of miles of cables used by the CYBERATTACKS: A NEW “KNOWN UNKNOWN” launch control centers to communicate with the silos and The increased frequency and sophistication of cyberattacks control the ICBMs. If the ICBMs were not on alert and ready on civilian and government systems in recent years has led to to be launched, they could not be launched by a cyberattacker. concerns about potential attacks on military warning systems Generals Cartwright and Dvorkin also note that cyberattacks or nuclear command and control systems, which could result on warning systems could result in false alarms (Cartwright in false alarms or unauthorized missile launches. Such cyber- and Dvorkin 2015). threats are not well understood and are continually evolving, significantly complicating efforts to defend against them. High Alert and Close Calls with A 2013 Defense Science Board study on advanced cyber- Nuclear Weapons attacks on military systems noted that “most of the systems” related to U.S. nuclear forces have not been fully assessed Historical incidents that involved false alarms in both the U.S. for vulnerability to cyberattack. It goes on to say that the U.S. and Russian systems remain relevant today for two reasons: military has “not kept up with the cyber adversary tactics and they illustrate the risks to any nuclear power of maintaining capabilities” and concludes that “With present capabilities missiles on high alert and thereby putting operators under and technology it is not possible to defend with confidence pressure to make fast decisions, and they illustrate the types against the most sophisticated cyber attacks” (Defense of incidents that can befall any complex human-engineered Science Board 2013). system. Indeed, military officers with experience overseeing Indeed, General Cartwright has written that, “Public U.S. nuclear forces are concerned about the vulnerability to reports of past experience with short time lines for decision cyberattack of missiles on high alert. The 2015 report by the making have shown that the process is flawed and that near

6 union of concerned scientists cataclysmic errors have been narrowly avoided but made more likely by the rushed nature of the process” (Cartwright 2013). Nuclear weapons systems are designed so that several things have to go wrong to result in an accidental, mistaken, or unauthorized missile launch. In past incidents only one or two things went wrong, so that in most cases the incident did not ultimately pose a serious risk. However, these incidents demonstrate that system failures do occur—and can occur in unexpected and compounding ways. Highlighted below are several incidents in both the United States and Russia that have increased the risk of nuclear disaster. They are part of a longer list of incidents in both countries that show the kinds of things that can—and do—go wrong (Union of Concerned Scientists 2015a; Schlosser 2013). Because military incidents resulting from mistakes are highly sensitive, there are likely more incidents that have not been made public and therefore remain unknown.

• False alarm due to warning sensors being fooled: On September 26, 1983, a Soviet early warning satellite reported five U.S. missiles attacking the Soviet Union. The alert came at a time of high tension between the two countries, due in part to the U.S. military buildup in the early 1980s and President ’s anti-Soviet rhetoric. In addition, earlier that month the Soviet Union shot down a Korean Airlines passenger plane that strayed into its airspace, killing nearly 300 people. The Soviet officer in command of the early warning center had only minutes to decide whether or not the satellite data was a false alarm. Since the satellite was found to be operating properly, following procedures would have led him to report an incoming attack. Going partly on gut instinct and also believing the United States was unlikely to launch an attack with only five missiles, he told his commanders that it was a false alarm before he actually knew that to be true. Later investigations revealed that the reflection of the sun on the tops of clouds had fooled the satellites into thinking it was detecting missile launches (Schlosser 2013; Hoffman 1999). In this case, high tensions between the two countries added significantly to the credibility of the warning. In the end, the short timeline did not allow enough time to determine that the warning was false before a decision was needed. In this case the strongest, and one of the few, safety links in the chain was the judgment of the officer

NASA in command of the early warning center. One important Early warning sensors can give accurate but ambiguous data that suggest an lesson: a different officer on duty that day may well have attack, such as in 1995 when the launch of a Norwegian scientific rocket (similar made a different—and tragic—decision. to the one above) was interpreted by Russian radar as a U.S. submarine-launched ballistic missile.

Reducing the Risk of Nuclear War 7 • False alarm due to accurate but ambiguous warning to assess the threat. The data showed all the signs of a data: On January 25, 1995, a Russian early warning radar nuclear attack. Within minutes, U.S. ICBM crews were detected a missile launch off the coast of Norway. Fearing notified, nuclear bombers prepared for takeoff, at least that it could be an SLBM launch intended to blind Rus- 10 fighter-interceptor planes were launched, and the sian radars as the first move in a larger attack, Russia president’s airborne command post (the National Emer- quickly put its nuclear forces on full alert. Fortunately, gency Airborne Command Post) took off (but without Russian satellites did not show any additional launches, the president). After six minutes, communication with and Russian leaders declared the incident a false alarm. U.S. radar sites indicated that the radars were not detect- What the radars had detected was actually the launch ing an attack, leading officials to decide no immediate of a Norwegian scientific rocket. Although Norway had action was necessary. Investigators later discovered the notified countries, including Russia, in advance of the incident was caused by a technician mistakenly inserting launch, the notification did not prevent a nuclear alert a training tape containing a scenario for a large-scale (Schlosser 2013). nuclear attack into an operational NORAD computer This example illustrates how coincidences can give (Sagan 1993). rise to false alarms. In this case, the upper stages of the Fortunately, tensions between the United States and sounding rocket coincidentally had similar speeds and Soviet Union were low in 1979, so there was some skep- altitudes to those of a U.S. Trident SLBM, so that a ticism about the warning. Within months, however, Russian computer identified it as a Trident missile. tensions spiked when the Soviets invaded Afghanistan, and continued to rise through President Reagan’s first term. Moreover, had communication systems been down On October 23, 2010, a or had the radars detected unrelated missile launches (or phenomena that appeared similar to missile launches), launch control center at the situation would have been much more serious.

Warren Air Force Base, • False alarm due to technical problems with the warn- WY, lost contact with the ing system: On June 3, 1980, U.S. early warning systems indicated a large incoming Soviet missile attack, trigger- 50 Minuteman III ICBMs ing a response by both SAC and other command centers. Bomber crews were ordered to their stations and started under its control. The their engines, and the National Emergency Airborne incident lasted for nearly Command Post prepared for a rapid takeoff. The alerts were suspended when warning systems showed no further an hour. The missiles were evidence of an attack. The U.S. Department of Defense on high alert and carrying later attributed the false alert to a failed computer chip. However, this problem created at least one more false nuclear warheads. alarm before it was identified (Comptroller General 1981). • Technical problems with command and control systems: On October 23, 2010, a launch control center Moreover, the launch location and path of the rocket at Warren Air Force Base, Wyoming, lost contact with the coincidentally made it appear to be well suited to blind- 50 Minuteman III ICBMs under its control. The incident, ing Russian radars. Fortunately, this false alarm hap- known as “launch facilities down,” lasted for nearly pened at a time of low tension between the two countries, an hour. The missiles were on high alert and carrying which led to skepticism that the launch was part of nuclear warheads. According to at least one report, there an attack. may have been previous communication problems at the • False alarm due to human error: On November 9, 1979, site that had not been corrected. A spokesperson said computers at the headquarters of NORAD (North American the launch control center was still able to monitor the Aerospace Defense Command) indicated a large-scale missiles but “We’ve never had something as big as this Soviet attack on the United States. NORAD relayed the happen . . . we’ve never lost complete command and con- information to Strategic Air Command (SAC) and other trol functionality of 50 ICBMs.” The cause of the problem high-level command posts, and top leaders convened was later found to be an electronic circuit card that had

8 union of concerned scientists been improperly installed in one of the computers the Watergate crisis. He was clinically depressed, during routine maintenance (Ambinder 2010). emotionally unstable, and drinking heavily. U.S. Secretary While much of the discussion of this latter incident of Defense James R. Schlesinger instructed the Joint focused on whether or not it had affected thereadiness Chiefs of Staff to route “any emergency order coming of the ICBMs, Bruce Blair, an analyst and former ICBM from the president”—such as a nuclear launch order— launch officer, noted that, “The more important concern through him first. Fortunately, this precaution was not should be that for the better part of an hour, the safe- needed (Schlosser 2013). guards that protect against unauthorized launch of These and other incidents make clear that unexpected things America’s missiles were compromised” since “the remote happen that increase the risk of a nuclear launch, and occur underground launch centers that control them lost their frequently enough to be a concern. And they show that events ability to detect and cancel any unauthorized launch that are considered highly unlikely, or are completely un- attempts” (Blair 2010). anticipated, do in fact occur. Indeed, the probability of an unintended catastrophic nuclear event may be larger than • Problems caused by personnel who did not follow is typically assumed. proper procedures: On August 29, 2007, six nuclear- armed cruise missiles were mistakenly loaded onto a B-52 bomber at Minot Air Force Base in North Dakota. Taking ICBMs Off High Alert Although there were multiple instances when the crew should have verified that the cruise missiles were not Taking an ICBM off high alert so that it can no longer be armed, no one followed required protocol to check for launched on a moment’s notice can be done in many different live weapons. The plane sat overnight on the tarmac at ways. How it is done would affect the amount of time it would Minot, unguarded. On August 30, it flew 1,500 miles to take to prepare the missile to launch and the ability of another a base in Louisiana where it sat unguarded for another country to verify that the missile is not on alert. nine hours until a maintenance crew realized that the Some options, such as removing the warheads from the weapons were live. In total, 36 hours passed before any- missiles and storing them separately, or putting a barrier over one in the U.S. Air Force realized that six live nuclear each silo that would have to be removed before the missile weapons were missing (Schlosser 2013). could be launched, could be relatively easy to verify, and would require a relatively long time to undo in preparation to The failure of Air Force personnel to know or follow launch. Other options, such as “safing” missiles by using the proper procedures raises concerns about whether similar safety switch in each silo that prevents the missile from being problems could exacerbate a crisis situation. In response launched when maintenance workers are in the silo, would to this incident, retired Air Force General Eugene Hab- be more difficult to verify, but could be done (and undone) iger, commander of U.S. Strategic Command from 1996 relatively quickly. to 1998, said, “I have been in the nuclear business since The option chosen could depend on whether the United 1966 and am not aware of any incident more disturbing” States removes its missiles from alert independently or in (Warrick and Pincus 2007). conjunction with Russia. Safing missiles could be the first • Increased risk due to an unstable individual high step in a process to negotiate a bilateral or multilateral agree- in the chain of command: In August 1974, President ment to keep missiles off high alert that would include Richard M. Nixon was in his last weeks in office during verification measures.

Taking ICBMs off high alert by safing has the advantage that it requires no additional equipment since it uses a switch that is part of all Minuteman III silos. Moreover, safing has been used in the past for this very purpose.

Reducing the Risk of Nuclear War 9 distributed over an area that is typically 10 to 20 miles on The failure of Air Force a side, with individual silos separated from one another by personnel to know or at least three miles. Each control center sits at a facility that includes a ground-level building housing six members of follow proper procedures a security force (Ahlborn et al. 2007). raises concerns about Maintenance workers visiting a silo complex enter through a hatch and go to the launcher equipment room. This whether similar problems room contains the electronic equipment that connects the missile to its launch control center and monitors the missile’s could exacerbate a crisis status. One of the first things the workers do is to activate the situation. safing switch that is part of this equipment; at that point the missile is manually safed (Union of Concerned Scientists 2015b). The switch opens a circuit that prevents a launch command from causing the missile’s first-stage motor to “SAFING” U.S. ICBMS ignite, but still allows the missile launch crew in the launch The launch crews for Minuteman III ICBMs are not located control center to monitor the missile’s status (Blair 2008). at the missile silos. Instead, two launch officers are stationed in a hardened, underground launch control center several miles from the silos. Each of these 45 centers monitors and controls a group or “flight” of 10 missiles. The 10 missiles are

Minuteman III Launch Facility

Electronic Launcher surveillance closure door system Personnel access hatch Launcher equipment room

Launcher equipment building

Launch eject gas generator U.S. Air Force

As this schematic shows, the launch facility is fully housed underground with a hatch at ground level that allows a crew Safing a Minuteman III is one option for taking these missiles off high alert. to enter the facility. The safing process, which is already routinely performed by maintenance work- ers, requires simply activating an existing switch in the launch control room. SOURCE: AHLBORN ET AL. 2007. This switch prevents the missile’s first-stage motor from igniting, and can eliminate the risk of mistaken launch.

10 union of concerned scientists Safing missiles in this way would eliminate the risk of mistaken launch and reduce the risk of accidental or unauthorized launch. Taking ICBMs off high alert by safing has the advantage that it requires no additional equipment since it uses a switch that is part of all Minuteman III silos. Moreover, safing has been used in the past for this very purpose: when President George H.W. Bush ordered 450 Minuteman II missiles imme- diately removed from high alert in 1991, it was done by safing the missiles using this switch. Estimates vary on how long it would take to safe all 450 U.S. ICBMs. Maintenance crews needed no more than two days to safe the 450 Minuteman II missiles in 1991 (SAC Office of the Historian 1991). However, Bruce Blair, who served as a missile control officer and has written extensively about this process, esti- mates all 450 missiles on alert today could be safed in as little as “about one-half day,” depending on the number of crews available (Blair 2008). Conversely, that is also the amount of time it could take to return the 450 safed missiles to alert fol- lowing a decision to do so. Safing the entire missile fleet could © Creative Commons/Char also be done over a longer period of time simply by having the maintenance crew leave the switch in the safe position after visiting a silo.

The launch control door from a Cold War–era Minuteman II launch site in Why the Arguments for Retaining South Dakota. High Alert Are Not Credible

Some administration officials and other advocates have States would not need to put its missiles back on alert in a advanced various arguments for maintaining rapid-launch crisis. Indeed, it is especially important that missiles be kept options. Below we show those arguments are not credible. off alert during a crisis, because that is when the risk of a mistaken launch would be greater. With options for launch A “RE-ALERTING RACE” COULD NOT OCCUR under attack removed, there would be no re-alerting in a crisis. With no re-alerting, there could be no re-alerting race. The primary reason the administration gives for not taking missiles off alert is that the process of putting them back on HIGH ALERT STATUS IS NOT NEEDED FOR DETERRENCE alert in the event of a crisis could lead to a “re-alerting race”— that is, a situation in which each side feels pressure to put its Deterrence is based on the concept that if a country launched missiles back on alert before the other can. Such a race could a nuclear attack on another country, it would still face the itself increase tensions and exacerbate a crisis (Department possibility of a devastating retaliatory nuclear strike. If that of State 2015a). Some argue it could even create a destabiliz- situation is clear to the potential attacker, then it would be ing window in which one country might think it could carry irrational for it to launch a first strike attack. out a first strike against the other. Because silo-based missiles are at known locations, they This argument is based on the fallacy that missiles on are vulnerable to a first strike. Keeping these missiles on high high alert and options for launching on warning are needed alert is intended to strengthen deterrence by making clear for deterrence. If that were true, it would require missiles they could be launched in retaliation before they could be to be put back on alert during a crisis. However, as discussed destroyed by a first strike. However, neither the United States below, missiles on high alert are not required for deterrence. nor Russia relies on silo-based missiles for deterrence. Instead, options to launch under attack should be com- The United States has spent many years developing its pletely removed from U.S. war plans. In that case the United submarine fleet to provide a secure second strike capability.

Reducing the Risk of Nuclear War 11 U.S. submarines are stealthy and cannot be detected at sea; they could not be targeted by Russia. The United States has The U.S. nuclear force is also put significant effort into making the command systems designed to be able to ride for these submarines survivable. Indeed, the administration recently stated, “Our command and control system and sub- out a first strike nuclear marines at sea would survive a surprise attack and enable a devastating response” (Department of State 2015a). Similarly, attack and still be able to a classified 2012 Pentagon study released in redacted form retaliate. states that Russia would not be able to achieve a significant military advantage “primarily because of the inherent surviv- ability of the planned U.S. strategic force structure, particu- on Japan, since such a response could lead to a nuclear larly the OHIO-class ballistic missile submarines, a number attack on U.S. territory.3 of which are at sea at any given time” (Office of the Secretary However, concerns about credibility do not depend on of Defense 2012). The U.S. nuclear force is designed to be able whether or not U.S. missiles are on alert; the issue is whether to ride out a first strike nuclear attack and be able to retaliate. the United States would be willing to launch a retaliatory Russia has developed both submarine-launched missiles strike, not how quickly it would do so. Moreover, recent UCS and mobile land-based missiles, which are less vulnerable discussions with a range of Japanese officials show that there than are silo-based missiles that sit at known locations. Such is broad support among the Japanese public and policy makers second strike forces on both sides would allow retaliation for having the United States and Russia take their missiles after incoming missiles land. off alert (Kulacki 2015b; Kulacki 2015c). Therefore, if Russia is rational and thus can be deterred, it will be, whether or not U.S. missiles are on high alert and HIGH ALERT STATUS IS NOT NEEDED FOR whether or not the president has the ability to launch on “NUCLEAR WARFIGHTING” warning. If it is irrational and cannot be deterred, then the In addition to the role nuclear weapons play in deterring high alert status of U.S. ICBMs is equally irrelevant. attacks, the U.S. nuclear war plan contains options for using Some might argue that the ability to launch on warning nuclear weapons in response to a nuclear attack, or in re- is just icing on the deterrence cake to make excessively clear sponse to a conventional, chemical, or biological attack by that the United States can retaliate following a first strike. If countries that have nuclear weapons. Using nuclear weapons there were no security costs associated with maintaining this in this way is typically called “nuclear warfighting” (Kris- option, then doing so might make sense. But that is not the tensen and McKinzie 2012). case, as the discussion above about the dangers of high alert Regardless of whether one believes such options make makes clear. sense, nuclear warfighting does not require having ICBMs on high alert. SLBMs have an accuracy comparable to or better HIGH ALERT STATUS IS NOT RELEVANT than that of ICBMs and could therefore be used to attack the TO EXTENDED DETERRENCE same targets (Cartwright 2013). These missiles would be Some analysts argue that changes to U.S. nuclear policy, available to launch if a decision was made to use them. such as taking missiles off alert, can be problematic because they may lead U.S. allies to question the U.S. commitment to THE “WARHEAD SPONGE” IS PRESERVED extended deterrence. Under extended deterrence, the United Some analysts argue that even if U.S. ICBMs are never States would respond to a nuclear attack on an ally with a launched, they are necessary to serve as what is sometimes retaliatory nuclear strike against the attacker. This policy is called a “warhead sponge”—meaning that Russia would need designed to deter such an attack in the first place. There is to devote a large fraction of its arsenal to targeting them if it ongoing tension over how to make a U.S. commitment to were considering a first strike, making it less likely to under- extended deterrence credible. For example, Japan has raised take such an attack. Regardless of the merits of this argument, concerns in the past about the credibility of the U.S. assurance the ICBMs would continue to serve as a warhead sponge as it would respond with nuclear weapons to a nuclear attack long as they are in their silos (with or without warheads)—

3 This situation is a key reason that the United States does not acknowledge nuclear vulnerability with respect to China, since such vulnerability is the basis of Japan’s concern.

12 union of concerned scientists their alert status is irrelevant. Any adversary planning a first that, “The current posture, which exerts pressure on the strike would still need to target them so that they could not President to make a nuclear choice rapidly, is a far greater be used in a retaliatory strike. So even off alert, U.S. ICBMs constraint [than removing launch under attack options]. would still act as a warhead sponge. Launch-under-attack pressure severely hobbles presidential decision making. It deprives our leaders of the time necessary REDUCING PRESIDENTIAL OPTIONS IS GOOD for deliberation and of the tools needed to direct U.S. power to coherent national purpose” (Cartwright 2013). Another argument sometimes made for keeping land- based missiles on alert is that all options should be kept available to the president, including options for launching BUREAUCRATIC POLITICS SHOULD NOT TRUMP NATIONAL SECURITY under attack. But if the risks of an option outweigh the benefits, One of the strongest motivations to keep missiles on alert then that option should not be on the table. Launching under appears to result from bureaucratic politics. The Air Force attack is not an option that either the U.S. or Russian presi- operates the U.S. ICBM force. Some analysts argue that if dent should have. silo-based missiles are taken off alert, it could be the first step Moreover, some military officials see the time constraints to eliminating such missiles altogether, which would get the imposed by retaining options to launch under attack them- Air Force out of the nuclear missile business (although it selves as needlessly and dangerously constraining to presi- would still oversee the nuclear bombers). There is, in fact, dential options. General Cartwright testified to the Senate ongoing discussion about whether the United States should © Steve Jurvetson/Creative Commons (Flickr) Jurvetson/Creative © Steve No longer part of the U.S. arsenal, the Titan II was a type of ICBM developed during the Cold War and kept on high alert.

Reducing the Risk of Nuclear War 13 eliminate its ICBMs and go from a triad of land-based missiles, also run a risk of unintentional strikes. The postures pose submarines, and bombers to a dyad of submarines and bomb- an existential threat to the very survival of the United ers. The accuracy of SLBMs has increased to the point that States, and Russia perceives no less cause for concern they can cover the same target set as ICBMs. (Cartwright 2013). Independent of the military utility of ICBMs, the Air Force might well fight to retain these missiles for bureau- In addition, Russia no longer has operational early warning cratic reasons alone, and push to keep the missiles on alert as satellites and thus relies exclusively on ground-based radars a way of implying their importance. Such parochial interests for warning (Podvig 2015). Such reliance significantly reduces should not be allowed to take precedence over a decision the amount of warning time Russia would receive of an in- to remove the missiles from high alert and thereby increase coming attack, which in turn reduces the time allowed for national and international security. assessment and a launch decision, further rushing these pro- cesses. It also eliminates the possibility of Russia gaining con- fidence in its warning data by receiving input from two The Evolving Strategic Environment independent sets of sensors (radars and satellites). Moreover, it is possible that China will decide to place its As a candidate and early in his first term, President Obama nuclear-armed missiles on alert for the first time. There is stated that policies to keep missiles on high alert are “unnec- now a discussion within Chinese military circles about this essary” and “increase the risk of catastrophic accidents or issue, which appears to stem from concerns on the part of the miscalculation,” and he pledged to “take our nuclear weapons military about retaining a credible nuclear retaliatory capabil- off hair-trigger alert” (Union of Concerned Scientists 2015c). ity in the face of U.S. missile defenses and accurate nuclear In completing its 2010 Nuclear Posture Review of U.S. nuclear weapons, and the development of high-precision convention- policies, however, the Obama administration decided to al weapons. The Chinese military appears to believe that put- maintain those rapid-launch options. ting its own missiles on high alert would be a step toward But the strategic environment has changed since 2010 ensuring its ability to retaliate against a nuclear attack, in in ways that require the administration to reassess this issue. addition to other steps it is taking including developing more Tensions between the United States and Russia are con- survivable mobile missiles and submarine-launched missiles siderably higher, which could increase the risk of a mistaken (Kulacki 2015a). launch, especially if the tensions escalate to a crisis. Indeed, in If China does decide to put nuclear missiles on high alert December 2015, former U.S. Secretary of Defense William Perry this would increase the risk of accidental, unauthorized, and declared, “today we now face the kind of dangers of a nuclear mistaken launches at the United States. The United States event like we had during the Cold War, an accidental war.” should therefore make it a priority to persuade China’s politi- He argued that the biggest threat is from ICBMs, which “are cal leaders, who will ultimately make this decision, to not put simply too easy to launch on bad information” (Mehta 2015). their weapons on alert. Since the policies of And General Cartwright has warned that: the United States are the most prominent external factor in- The current launch-ready postures of the United States fluencing Chinese advocates of raising the alert level of Chi- and Russia are major sources of instability. They not only na’s nuclear forces, an announcement that the United States would generate pressure on leaders to make a premature is ending its own practice of keeping missiles on high alert decision on the use of nuclear weapons in a crisis, but they could have an impact on the Chinese debate.

If China decides to put its nuclear missiles on high alert this would increase the risk of accidental, unauthorized, and mistaken launches at the United States. The United States could help persuade China’s political leaders not to take this step by ending its own practice of keeping missiles on high alert.

14 union of concerned scientists This report shows that the risk of accidental, unauthor- Defense Science Board. 2013. Resilient military systems and the ized, or mistaken launch of nuclear missiles is real, and that advanced cyber threat. January. Online at www.acq.osd.mil/dsb/ reports/ResilientMilitarySystems.CyberThreat.pdf. there is no reason for the United States to continue to accept Department of Defense. 1981. Narrative summaries of accidents this risk. By taking its land-based nuclear missiles off high involving U.S. nuclear weapons 1950–1980. April. Online at alert and removing rapid-launch options from its nuclear https://nsarchive.files.wordpress.com/2010/04/635.pdf. war plans, the United States would enhance both U.S. and Department of State. 2015a. U.S. nuclear force posture and de-alerting. Fact sheet. December 14. Online at www.state.gov/t/avc/ international security. rls/250644.htm. Department of State. 2015b. Nuclear force policy and posture. U.S. national report to the Nuclear Non-Proliferation Treaty David Wright is a co-director of the UCS Global Security review conference. Online at www.state.gov/documents/ Program. Eryn MacDonald is an analyst for the program. organization/242083.pdf. Lisbeth Gronlund is a co-director of the program. Feynman, R.P. 1986. Personal observations on reliability of shuttle. Report of the Presidential Commission on the Space Shuttle Challenger Accident. Appendix F. June 6. Online at http://history. ACKNOWLEDGMENTS nasa.gov/rogersrep/v2appf.htm. This report was made possible by the generous support of The Prospect Hill Frontline. 1999a. Interview: General William Odom (ret.). Public Foundation, Telemachus: Foundation to Empower the Poor and End War, Broadcasting System. February 1999. Online at www.pbs.org/ and members of the Union of Concerned Scientists. wgbh/pages/frontline/shows/russia/interviews/odom.html. We are grateful to Bryan Wadsworth and Cynthia DeRocco for their help Frontline. 1999b. Interview: Admiral Stansfield Turner. Public in coordinating the production, to Trudy Bell for her editing skills, and to David Gerratt for layout and design expertise. Broadcasting System. February 1999. Online at www.pbs.org/ The opinions expressed herein do not necessarily reflect those of the organiza- wgbh/pages/frontline/shows/russia/interviews/turner.html. tions that funded the work or the individuals who reviewed it. The Union of Global Zero. 2015. De-alerting and stabilizing the world’s nuclear force Concerned Scientists bears sole responsibility for the report’s content. postures. April. Online at www.globalzero.org/files/global_zero_ commission_on_nuclear_risk_reduction_report_0.pdf. REFERENCES Global Zero. 2012. Global Zero U.S. Nuclear Policy Commission report: Ahlborn, D., D. Coleman, S. Hogan-Conrad, D. Jury, and M. Malle. Modernizing U.S. nuclear strategy, force structure and posture. 2007. Environmental assessment: Minuteman III deactivation, May. Online at www.globalzero.org/files/gz_us_nuclear_policy_ Malmstrom AFB, Montana. Great Falls, MT: Malmstrom Air Force commission_report.pdf. Base. Online at www.dtic.mil/dtic/tr/fulltext/u2/a496029.pdf. Habiger, E. 2002. Testimony before the U.S. Senate Committee on Ambinder, M. 2010. Failure shuts down squadron of nuclear missiles. Foreign Relations. July 23. Online at www.gpo.gov/fdsys/pkg/ The Atlantic. October 26. Online at www.theatlantic.com/politics/ CHRG-107shrg81339/pdf/CHRG-107shrg81339.pdf. archive/2010/10/failure-shuts-down-squadron-of-nuclear- Hoffman, D. 1999. “I had a funny feeling in my gut.”Washington Post, missiles/65207/. February 10. Online at www.washingtonpost.com/wp-srv/inatl/ Blair, B.G. 2015. Could U.S.–Russia tensions go nuclear? Politico. longterm/coldwar/shatter021099b.htm. November 27. Online at www.politico.com/magazine/ Kristensen, H.M, and M. McKinzie. 2012. Reducing alert rates of story/2015/11/russia-us-tensions-nuclear-cold-war- nuclear weapons. Geneva: United Nations Institute for Disarm- 213395#ixzz3siPwmy7c. ament Research. UNIDIR/2012/6. Online at www.unidir.org/files/ Blair, B.G. 2010. Could terrorists launch America’s nuclear missiles? publications/pdfs/reducing-alert-rates-of-nuclear-weapons-400. Time. November 11. Online at http://content.time.com/time/ pdf. nation/article/0,8599,2030685,00.html. Kulacki, G. 2015a. China’s military calls for putting its nuclear forces Blair, B.G. 2008. De-alerting strategic forces. In Reykjavik revisited: on alert. Cambridge, MA: Union of Concerned Scientists. Steps toward a world free of nuclear weapons, edited by G.P. December. Online at www.ucsusa.org/ChinaHairTrigger. Shultz, S.D. Drell, S.P. Andreasen, and J.E. Goodby. Stanford, CA: Kulacki, G. 2015b. Japan’s top hawk calls for the United States to end Hoover Institution Press. Online at http://media.hoover.org/sites/ hair-trigger alert. Blog. Cambridge, MA: Union of Concerned default/files/documents/9780817949211_ch2.pdf. Scientists. April 27. Online at http://blog.ucsusa.org/gregory- Cartwright, J.E. 2013. Examining the proper size of the nuclear kulacki/japan-abe-hair-trigger-alert-716. weapons stockpile to maintain a credible U.S. deterrent. Prepared Kulacki, G. 2015c. Japan calls for United States to end hair-trigger testimony to the United States Senate Committee on Appropriations alert. Cambirdge, MA: Union of Concerned Scientists. April. Sub-committee on Energy and Water Development. July 25. Online at www.ucsusa.org/sites/default/files/attach/2015/04/ Online at www.appropriations.senate.gov/imo/media/doc/ Japan-hair-trigger-fact-sheet.pdf. hearings/07_25_12%20E&W%20Nuclear%20Weapons%20 Stockpile%20as%20Deterrent%20GPO%20record.pdf. McNamara, R.S. 2009. Apocalypse soon. Foreign Policy. May/June. Online at http://foreignpolicy.com/2009/10/21/apocalypse-soon. Cartwright, J.E., and V. Dvorkin. 2015. How to avert a nuclear war. New York Times, April 19. Op-ed. Online at www.nytimes. Mehta, A. 2015. Former SecDef Perry: US on “brink” of new nuclear com/2015/04/20/opinion/how-to-avert-a-nuclear-war. arms race. Defense News, December 3. Online at www. html?partner=rss&emc=rss&smid=tw-nytimesworld&_r=0. defensenews.com/story/defense/policy-budget/2015/12/03/ former-secdef-perry-us-brink-new-nuclear-arms-race/76721640. Comptroller General of the United States. 1981. NORAD’s missile warning system: What went wrong? Report to the chairman. Nunn, S. 2002. Testimony before the U.S. Senate Committee on Committee on Government Operations. House of Representa- Foreign Relations. July 23. Online at www.gpo.gov/fdsys/pkg/ tives. MASAD-81-30. U.S. General Accounting Office. May 15. CHRG-107shrg81339/pdf/CHRG-107shrg81339.pdf. Online at www.gao.gov/assets/140/133240.pdf.

Reducing the Risk of Nuclear War 15 Office of the Secretary of Defense. 2012.Report on the strategic nuclear Strategic Air Command (SAC), Office of the Historian. 1991.Alert forces of the Russian Federation pursuant to Section 1240 of the operations and the Strategic Air Command, 1957–1991. Offutt Air Force National Defense Authorization Act for Fiscal Year 2012. U.S. Base, NE: Headquarters, Strategic Air Command. Online Department of Defense. Partially declassified and released under the at https://books.google.com/books?id=Y8jh37NJf3YC. Freedom of Information Act. Online at http://fas.org/programs/ssp/ Union of Concerned Scientists. 2015a. Close calls with nuclear weapons. nukes/nuclearweapons/DOD2012_RussianNukes.pdf. Fact sheet. Online at www.ucsusa.org/sites/default/files/ Office of the Secretary of Defense. 2010. Nuclear posture review report. attach/2015/04/Close%20Calls%20with%20Nuclear%20Weapons.pdf. U.S. Department of Defense. April. Online at www.defense.gov/ Union of Concerned Scientists. 2015b. A simple method for taking U.S. Portals/1/features/defenseReviews/NPR/2010_Nuclear_Posture_ land-based nuclear missiles off high alert. Fact sheet. Online at www. Review_Report.pdf. ucsusa.org/sites/default/files/attach/2015/04/safing-us-nuclear- Perrow, C. 1984. Normal accidents: Living with high-risk technologies. missiles.pdf. New York, NY: Basic Books. Union of Concerned Scientists. 2015c. U.S. military and political leaders Podvig, P. 2015. Early warning. Russian Strategic Nuclear Forces, urge taking nuclear weapons off hair-trigger alert.Fact sheet. Online February 11. Online at http://russianforces.org/sprn. at www.ucsusa.org/sites/default/files/attach/2015/01/leaders-against- Podvig, P. (editor). 2001. Russian strategic nuclear forces. Cambridge, hair-trigger-alert.pdf. MA: The MIT Press. Warrick, J., and W. Pincus. 2007. Missteps in the bunker. Washington Sagan, S.D. 1993. The limits of safety: Organizations, accidents, and Post, September 23. Online at www.washingtonpost.com/wp-dyn/ nuclear weapons. Princeton: Princeton University Press. content/article/2007/09/22/AR2007092201447.html. Schlosser, E. 2013. Command and control. New York: Penguin Press. Starr, S. No date. Launch-ready nuclear weapons: A threat to all nations and peoples. Physicians for Social Responsibility. Online at www.psr.org/nuclear-weapons/launch-ready-nuclear-weapons.pdf.

find this document online: www.ucsusa.org/ReducingtheRisk

The Union of Concerned Scientists puts rigorous, independent science to work to solve our planet’s most pressing problems. Joining with citizens across the country, we combine technical analysis and effective advocacy to create innovative, practical solutions for a healthy, safe, and sustainable future.

NATIONAL HEADQUARTERS WASHINGTON, DC, OFFICE WEST COAST OFFICE MIDWEST OFFICE Two Brattle Square 1825 K St. NW, Suite 800 500 12th St., Suite 340 One N. LaSalle St., Suite 1904 Cambridge, MA 02138-3780 Washington, DC 20006-1232 Oakland, CA 94607-4087 Chicago, IL 60602-4064 Phone: (617) 547-5552 Phone: (202) 223-6133 Phone: (510) 843-1872 Phone: (312) 578-1750 Fax: (617) 864-9405 Fax: (202) 223-6162 Fax: (510) 843-3785 Fax: (312) 578-1751

web: www.ucsusa.org printed on recycled paper using vegetable-based inks © JANUARY 2016 union of concerned scientists